OSDev.org

I always wondered of the purpose of monitor drivers until I started reading about colour space.

sRGB (http://en.wikipedia.org/wiki/SRGB) only shows a small portion of the possible visible colours. Depending on your monitors brightness/contrast/temperature etc the sRGB triangle can be mapped do a different area of the spectrum. So it made sense that for professional photography you would need monitor/printer drivers that can reproduce the exact some colour from one device onto another.

http://en.wikipedia.org/wiki/ProPhoto_RGB_color_space comes close to representing the fuller range but not quite. I've looked into other systems aswell (L.a.b being my favourite).

So I'm thinking about making a colour model that can represent the entire range of visible colours (even if it touches upon ultraviolet). I thought about a physics based model based on photon wave length and intensity, though some colours (e.g. pink) are made up of multiple waves. I'm looking for a system that wouldn't just be for storing images, but also practical to do lighting operations on (e.g. with RGB it's easy to multiply (reflected colour = surface colour * light colour) and add light (output = light 1 + light 2 + light 3) together in real time).

Kind of off-topic, but this is why I like BenQ monitors; they provide their monitor drivers on Windows Update. So (assuming you use Windows) you get the drivers automatically without noticing. Definitely bonus points there.

Hi,

Looking at the CIE chromaticity diagram, in order to incorporate the entire visible spectrum without too much outside CIE colour space, you'dhave to introduce several more coordinates. IIRC, there was an attempt to create such a system with a s dimensional variant of the chromaticity diagram.

At uni, I had several lighting and colour modules (relating to the human eye, not computers) - I'll see if I can dig out any of my old notes over the next few days.

Cheers,
Adam

JackScott wrote:Kind of off-topic, but this is why I like BenQ monitors; they provide their monitor drivers on Windows Update. So (assuming you use Windows) you get the drivers automatically without noticing. Definitely bonus points there.

A lot of monitor and device vendors do this now. Last time I reinstalled Windows, I didn't have to download a single driver.

So I'm thinking about making a colour model that can represent the entire range of visible colours

How will you represent gold?

Craze Frog wrote:How will you represent gold?

Thats easy from a colour point of view. What's slightly more tricky is the reflectivity and smooth texture of polished gold, which can be simulated easily enough with current hardware.

Cheers,
Adam

AJ wrote:

Craze Frog wrote:How will you represent gold?

Thats easy from a colour point of view.

Cheers,
Adam

Yellow and gold look different. Would you encode them as if they were the same?

The point of color spaces is that the human eye cannot SEE all possible colors distinctly. So it makes no sense at all to try to reproduce them, physically. (And, of course, each human's eyes are slightly different, and respond differently to the same mixture of wavelengths and intensities.)

A simple example: create a laser that produces a precise 5555nm wavelength. Ask people what they see. They will say "green". Create a careful mixture of RGB intensities, and ask people if it is the same color. They will say "yes". Will the mixture be 0 red, and 0 blue? No, it will not. (Especially since the "green" in the RGB is not at 5555nm.) So, question: what, physically, does that mixture of RGB have to do with 5555nm light? Answer: nothing whatsoever. The only point is that human eyes happen to see 5555nm light, and that mixture of RGB, and billions of other mixtures of billions of other wavelengths and intensities as being basically the same color.

Now, yes, it is true that most humans can experience colors that RGB or any other colorspace cannot quite reproduce. It is also true that blue-eyed humans can see soft ultraviolet, where brown-eyed humans cannot.

But the point is going to be that you are dealing with a marginal effect, here. And that on that margin, individual human differences become much more of a factor. All the current color schemes fairly accurately represent the majority of colors that everyone can agree on.

And yes, humans recognize things like "gold" by the observed color and intensity of the specular reflections, not by the precise wavelengths.

Hi,

Hopefully in the post above, you mean 555nm

As an addendum to the above, the CIE diagram takes in to account different percieved colours being produced by different mixes - pick two points, draw a line joining them. The point half way along that line shows the colour created by the two original primaries.

Another interesting point is the straight "line of purples" along the bottom of the diagram. It represents colours which cannot be obtained by a single wavelength of light - they have to be mixed.

Cheers,
Adam

Meant Angstroms, actually -- guess I was tired.

Craze Frog wrote:

AJ wrote:

Craze Frog wrote:How will you represent gold?

Thats easy from a colour point of view.

Cheers,
Adam

Yellow and gold look different. Would you encode them as if they were the same?

Yes and no - if the light hitting waves hitting the eyes photo receptors in these two cases "mapped" to the same "color value" then yes - otherwise no.

How the human visual system perceives colors is very complex (and not fully understood) but can basically be modeled with 3 "primary" colors since we have (roughly) 3 kinds of light receptors in the eye. This means that any combination of base colors can be used and only three parameters is enough to model the colors that we can see. Different color systems can be viewed as different mappings of the same.

The above is a very simplified explanation and is I probably made some mistakes - please correct me.

clange

bewing wrote:All the current color schemes fairly accurately represent the majority of colors that everyone can agree on.

That's simply not true. I learned that from experience after getting education in arts. There is a very significant part of RGB that can't be represented properly with CMYK, for example. I'm not talking about minor visual differences here, I'm talking about huge ones.

clange wrote:This means that any combination of base colors can be used and only three parameters is enough to model the colors that we can see. Different color systems can be viewed as different mappings of the same.

The above is a very simplified explanation and is I probably made some mistakes - please correct me.

Wikipedia: Representation of gamuts is a very good paragraph to de-illusion you.

Solar wrote:

clange wrote:This means that any combination of base colors can be used and only three parameters is enough to model the colors that we can see. Different color systems can be viewed as different mappings of the same.

The above is a very simplified explanation and is I probably made some mistakes - please correct me.

Wikipedia: Representation of gamuts is a very good paragraph to de-illusion you.

As I said: very simplified. It is of course only true with respect to the human visual system (*) and under the assumption that the color spaces are not constraint by reality. The color space as such does not restrict the which colors can be described only the gamut of the device used to render the final color. Which also gives a good intuitive explanation for negative color coefficients.

(*) And three samples is of course not enough to describe colors when doing calculations. Imagine a fluorescent light source illuminating a lemon. 3 samples will never be enough to represent both the relative smooth "reflection" of the lemon compared to the very narrow spike of the light source.

clange wrote:(*) And three samples is of course not enough to describe colors when doing calculations. Imagine a fluorescent light source illuminating a lemon. 3 samples will never be enough to represent both the relative smooth "reflection" of the lemon compared to the very narrow spike of the light source.

Erm...

I think you are confusing color (which could be a single pixel on the screen) and stuff like texture, reflection etc...